wav2vec2-common_voice-en-finetune

This model is a fine-tuned version of facebook/wav2vec2-large-xlsr-53 on the GOOGLE/FLEURS - EN_US dataset. It achieves the following results on the evaluation set:

• Loss: 0.3436
• Wer: 0.2598

Model description

More information needed

Intended uses & limitations

More information needed

Training and evaluation data

More information needed

Training procedure

Training hyperparameters

The following hyperparameters were used during training:

• learning_rate: 0.0003
• train_batch_size: 16
• eval_batch_size: 8
• seed: 42
• gradient_accumulation_steps: 2
• total_train_batch_size: 32
• optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
• lr_scheduler_type: linear
• lr_scheduler_warmup_steps: 500
• num_epochs: 10.0
• mixed_precision_training: Native AMP

Training results

Training Loss	Epoch	Step	Validation Loss	Wer
No log	1.0870	100	3.5106	1.0
No log	2.1739	200	2.9226	1.0
No log	3.2609	300	2.8745	1.0
No log	4.3478	400	1.8100	0.9804
3.7609	5.4348	500	0.4771	0.4207
3.7609	6.5217	600	0.3808	0.3484
3.7609	7.6087	700	0.3408	0.2872
3.7609	8.6957	800	0.3479	0.2719
3.7609	9.7826	900	0.3437	0.2604

Framework versions

• Transformers 4.45.0.dev0
• Pytorch 2.4.0+cu121
• Datasets 2.21.0
• Tokenizers 0.19.1